Dataframe datamodule

CHANGELOG.md

@@ -8,6 +8,8 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
 
 ### Added
 
+- 🚀 Add Dataframe datamodule by @manuelkonrad in https://github.com/openvinotoolkit/anomalib/pull/2403
+
 ### Changed
 
 ### Deprecated

configs/data/dataframe.yaml

@@ -0,0 +1,77 @@
+class_path: anomalib.data.Dataframe
+init_args:
+  name: bottle
+  root: "datasets/MVTec/bottle"
+  train_batch_size: 32
+  eval_batch_size: 32
+  num_workers: 8
+  task: segmentation
+  transform: null
+  train_transform: null
+  eval_transform: null
+  test_split_mode: from_dir
+  test_split_ratio: 0.2
+  val_split_mode: same_as_test
+  val_split_ratio: 0.5
+  seed: null
+  samples:
+    - image_path: train/good/000.png
+      label_index: 0
+      mask_path: ""
+      split: train
+    - image_path: train/good/001.png
+      label_index: 0
+      mask_path: ""
+      split: train
+    - image_path: train/good/002.png
+      label_index: 0
+      mask_path: ""
+      split: train
+    - image_path: train/good/003.png
+      label_index: 0
+      mask_path: ""
+      split: train
+    - image_path: train/good/004.png
+      label_index: 0
+      mask_path: ""
+      split: train
+    - image_path: test/bad/000.png
+      label_index: 1
+      mask_path: ground_truth/bad/000_mask.png
+      split: test
+    - image_path: test/bad/002.png
+      label_index: 1
+      mask_path: ground_truth/bad/002_mask.png
+      split: test
+    - image_path: test/bad/004.png
+      label_index: 1
+      mask_path: ground_truth/bad/004_mask.png
+      split: test
+    - image_path: test/good/000.png
+      label_index: 0
+      mask_path: ""
+      split: test
+    - image_path: test/good/001.png
+      label_index: 0
+      mask_path: ""
+      split: test
+    - image_path: test/good/003.png
+      label_index: 0
+      mask_path: ""
+      split: test
+    - image_path: test/bad/001.png
+      label_index: 1
+      mask_path: ground_truth/bad/001_mask.png
+      split: test
+    - image_path: test/bad/003.png
+      label_index: 1
+      mask_path: ground_truth/bad/003_mask.png
+      split: test
+    - image_path: test/good/002.png
+      label_index: 0
+      mask_path: ""
+      split: test
+    - image_path: test/good/004.png
+      label_index: 0
+      mask_path: ""
+      split: test

docs/source/markdown/guides/reference/data/image/dataframe.md

@@ -0,0 +1,7 @@
+# Dataframe Data
+
+```{eval-rst}
+.. automodule:: anomalib.data.image.dataframe
+   :members:
+   :show-inheritance:
+```

docs/source/markdown/guides/reference/data/image/index.md

@@ -16,6 +16,13 @@ Learn more about BTech dataset.
 Learn more about custom folder dataset.
 :::
 
+:::{grid-item-card} Dataframe
+:link: ./dataframe
+:link-type: doc
+
+Learn more about custom dataframe dataset.
+:::
+
 :::{grid-item-card} Kolektor
 :link: ./kolektor
 :link-type: doc

src/anomalib/data/__init__.py

@@ -14,7 +14,7 @@
 
 from .base import AnomalibDataModule, AnomalibDataset
 from .depth import DepthDataFormat, Folder3D, MVTec3D
-from .image import BTech, Datumaro, Folder, ImageDataFormat, Kolektor, MVTec, Visa
+from .image import BTech, Dataframe, Datumaro, Folder, ImageDataFormat, Kolektor, MVTec, Visa
 from .predict import PredictDataset
 from .utils import LabelName
 from .video import Avenue, ShanghaiTech, UCSDped, VideoDataFormat
@@ -70,6 +70,7 @@ def get_datamodule(config: DictConfig | ListConfig | dict) -> AnomalibDataModule
     "VideoDataFormat",
     "get_datamodule",
     "BTech",
+    "Dataframe",
     "Datumaro",
     "Folder",
     "Folder3D",

src/anomalib/data/image/__init__.py

@@ -9,6 +9,7 @@
 from enum import Enum
 
 from .btech import BTech
+from .dataframe import Dataframe
 from .datumaro import Datumaro
 from .folder import Folder
 from .kolektor import Kolektor
@@ -20,6 +21,7 @@ class ImageDataFormat(str, Enum):
     """Supported Image Dataset Types."""
 
     BTECH = "btech"
+    DATAFRAME = "dataframe"
     DATUMARO = "datumaro"
     FOLDER = "folder"
     FOLDER_3D = "folder_3d"
@@ -29,4 +31,4 @@ class ImageDataFormat(str, Enum):
     VISA = "visa"
 
 
-__all__ = ["BTech", "Datumaro", "Folder", "Kolektor", "MVTec", "Visa"]
+__all__ = ["BTech", "Dataframe", "Datumaro", "Folder", "Kolektor", "MVTec", "Visa"]